sage-5.10.beta5 build error for scipy-0.12.0.p0.log

29 views
Skip to first unread message

Stephen Montgomery-Smith

unread,
May 27, 2013, 3:52:59 PM5/27/13
to sage-...@googlegroups.com
This is building with the FreeBSD port, using SAGE_ATLAS_LIB=/usr/local/lib.

I really don't know what to look for, or which errors are fatal and
which are just warnings.

See attachment.
scipy-0.12.0.p0.log

Jean-Pierre Flori

unread,
May 27, 2013, 4:55:31 PM5/27/13
to sage-...@googlegroups.com
Heyn
Could you post the atlas log as well please?

Best,
JP

Stephen Montgomery-Smith

unread,
May 27, 2013, 4:59:12 PM5/27/13
to sage-...@googlegroups.com
Here it is.

atlas-3.10.1.p0.log

Jean-Pierre Flori

unread,
May 27, 2013, 5:06:14 PM5/27/13
to sage-...@googlegroups.com
And config.log from scipy build folder?

Stephen Montgomery-Smith

unread,
May 27, 2013, 5:17:08 PM5/27/13
to sage-...@googlegroups.com
There isn't one. It looks like scipy uses setup.py instead of configure.
> --
> You received this message because you are subscribed to the Google
> Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sage-devel?hl=en.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>

Jean-Pierre Flori

unread,
May 27, 2013, 5:27:58 PM5/27/13
to sage-...@googlegroups.com
IIRC scipy adds strange prefix to atlas library names on FreeBSD (_r things).
I seem to remember struggling with that on my previous FreeBSD attempt.
Not sure though why I did not stumble upon that durng my current attempt.

Just checked and as I used a mixture of 5.9 + #10508 without removing the (reference) blas and lapack spkg, it seems scipy found the libraries built by the blas and lapack spkgs.

Jean-Pierre Flori

unread,
May 27, 2013, 5:32:20 PM5/27/13
to sage-...@googlegroups.com


On Monday, May 27, 2013 11:27:58 PM UTC+2, Jean-Pierre Flori wrote:
IIRC scipy adds strange prefix to atlas library names on FreeBSD (_r things).
I seem to remember struggling with that on my previous FreeBSD attempt.
Not sure though why I did not stumble upon that durng my current attempt.

Just checked and as I used a mixture of 5.9 + #10508 without removing the (reference) blas and lapack spkg, it seems scipy found the libraries built by the blas and lapack spkgs.
In fact it seems that it is numpy which is tricky.
See my own rants at https://groups.google.com/d/msg/sage-devel/b8ZGqE6Y7Vg/XLkduDIJsBUJ

Jean-Pierre Flori

unread,
May 27, 2013, 5:36:16 PM5/27/13
to sage-...@googlegroups.com

Stephen Montgomery-Smith

unread,
May 27, 2013, 5:42:57 PM5/27/13
to sage-...@googlegroups.com
I just found out that beta5 has atlas version 3.10.1, and no spkg for
blas, whereas beta4 used atlas 3.8.4 and has a spkg for blas. It looks
like your patch is applied to beta5.

So I will try rebuilding without SAGE_ATLAS_LIB set.
> > an email to sage-devel+...@googlegroups.com <javascript:>.
> > To post to this group, send email to sage-...@googlegroups.com
> <javascript:>.
> <http://groups.google.com/group/sage-devel?hl=en>.
> > For more options, visit https://groups.google.com/groups/opt_out
> <https://groups.google.com/groups/opt_out>.

Stephen Montgomery-Smith

unread,
May 27, 2013, 5:49:04 PM5/27/13
to sage-...@googlegroups.com
Our emails crossed. I'll try this one first.

Jean-Pierre Flori

unread,
May 27, 2013, 5:55:58 PM5/27/13
to sage-...@googlegroups.com
Yup, I think ATLAS was in fact never used on FreeBSD, only the reference blas and lapack.
But now that we do not ship them anymore, we'll have to patch numpy to help scipy (?!!?!) to find ATLAS...

Stephen Montgomery-Smith

unread,
May 27, 2013, 7:12:37 PM5/27/13
to sage-...@googlegroups.com
It worked!

Thank you very much.

Jean-Pierre Flori

unread,
May 28, 2013, 4:30:45 AM5/28/13
to sage-...@googlegroups.com
Reply all
Reply to author
Forward
0 new messages